Proc. Natl. Acad. Sci. USAVol. 76, No. 9, pp. 4350-4354, September 1979Biochemistry

Electrophoretic transfer of proteins from polyacrylamide gels tonitrocellulose sheets: Procedure and some applications

(ribosomal proteins/radioimmunoassay/fluorescent antibody assay/peroxidase-conjugated antibody/autoradiography)

HARRY ToWBIN*, THEOPHIL STAEHELINt, AND JULIAN GORDON*t*Friedrich Miescher-Institut, P. 0. Box 273, CH-4002 Basel, Switzerland; and tPharmaceutical Research Department, Hoffman-La Roche,CH-4002 Basel, Switzerland

Communicated by V. Prelog, June 12, 1979

ABSTRACT A method has been devised for the electro-phoretic transfer of proteins from polyacrylamide gels to ni-trocellulose sheets. The method results in quantitative transferof ribosomal proteins from gels containing urea. For sodiumdodecyl sulfate gels, the original band pattern was obtainedwith no loss of resolution, but the transfer was not quantitative.The method allows detection of proteins by autoradiographyand is simpler than conventional procedures. The immobilizedproteins were detectable by immunological procedures. Alladditional binding capacity on the nitrocellulose was blockedwith excess protein; then a specific antibody was bound and,finally, a second antibody directed against the first antibody.The second antibody was either radioactively labeled or con-jugated to fluorescein or to peroxidase. The specific protein wasthen detected by either autoradiography, under UV light, or bythe peroxidase reaction product, respectively. In the latter case,as little as 100 pg of protein was clearly detectable. It is antici-pated that the procedure will be applicable to analysis of a widevariety of proteins with specific reactions or ligands.

Polyacrylamide gel electrophoresis has become a standard toolin every laboratory in which proteins are analyzed and purified.Most frequently, the amount and location of the protein are ofinterest and staining is then sufficient. However, it may also beimportant to correlate an activity of a protein with a particularband on the gel. Enzymatic and binding activities can some-times be detected in situ by letting substrates or ligands diffuseinto the gel (1, 2). In immunoelectrophoresis, the antigen isallowed to diffuse (3) or electrophoretically move (4) againstantibody. A precipitate is then formed where the antigen andantibody interact. Modifications have been described in whichthe antigen is precipitated by directly soaking the separationmatrix in antiserum (5, 6). The range of gel electrophoreticseparation systems is limited by the pore size of the gels anddiffusion of the antibody. The systems are also dependent onconcentration and type of antigen or antibody to give a physi-cally immobile aggregate.

Analysis of cloned DNA has been revolutionized (7) by theability to fractionate the DNA electrophoretically in polyac-rylamide/agarose gels first and then to obtain a faithful replicaof the original gel pattern by blotting the DNA onto a sheet ofnitrocellulose on which it is immobilized. The immobilizedDNA can then be analyzed by in situ hybridization. The powerof immobilized two-dimensional arrays has been extended tothe analysis of proteins by use of antibody-coated plastic sheetsto pick up the corresponding antigen from colonies on agarplates (8). Sharon et al. (9) have used antigen-coated nitrocel-lulose sheets to pick up antibodies secreted by hybridoma clonesgrowing in agar.

In this report we describe a procedure for the transfer of

proteins from a polyacrylamide gel to a sheet of nitrocellulosein such a way that a faithful replica of the original gel patternis obtained. A wide variety of analytical procedures can beapplied to the immobilized protein. Thus, the extreme versa-tility of nitrocellulose binding assays can be combined withhigh-resolution polyacrylamide gel electrophoresis. The pro-cedure brings to the analysis of proteins the power that theSouthern (7) technique has brought to the analysis of DNA.

MATERIALS AND METHODSImmunogens and Immunization Procedures. Escherichia

coli ribosomal proteins L7 and L12 were extracted (10) from50S subunits and purified as described (11) by ion-exchangechromatography on carboxymethyl- and DEAE-cellulose.Antibodies were raised in a goat by injecting 250 jig of proteinemulsified with complete Freund's adjuvant intracutaneouslydistributed over several sites. Bacillus pertussis vaccine (1.5 mlof Bordet-Gengou vaccine, Schweizerisches Serum- undImpfinstitut, Bern, Switzerland) was given subcutaneously withevery antigen injection. Booster injections of the same formu-lation were given on days 38, 79, and 110. The animal was bledon day 117.

Subunits from chicken liver ribosomes (12) were combinedin equimolar amounts, and 200-iAg aliquots were emulsifiedwith 125 Al of complete Freund's adjuvant injected at one in-traperitoneal and four subcutaneous sites into BALB/c mice.Booster injections of 400 ,ig of ribosomes in saline were givenintraperitoneally on days 33, 57, 58, and 59. The animals werebled on day 71.

Electrophoretic Blotting Procedures. Proteins were firstsubjected to electrophoresis in the presence of urea either in twodimensions (12) or in one-dimensional slab gels correspondingto the second dimension of the same two-dimensional system.The proteins were then transferred to nitrocellulose sheets asfollows. The physical assembly used is shown diagrammaticallyin Fig. 1. A sheet of nitrocellulose (0.45 ,im pore size in rollform, Millipore) was briefly wetted with water and laid on ascouring pad (Scotch-Brite) which was supported by a stiffplastic grid (disposable micropipette tray, Medical LaboratoryAutomation, Inc., New York). The gel to be blotted was put onthe nitrocellulose sheet and care was taken to remove all airbubbles. A second pad and plastic grid were added and rubberbands were strung around all layers. The gel was thus firmlyand evenly pressed against the nitrocellulose sheet. The as-sembly was put into an electrophoretic destaining chamber withthe nitrocellulose sheet facing the cathode. The chamber con-tained 0.7% acetic acid. A voltage gradient of 6 V/cm was ap-plied for 1 hr.

For polyacrylamide electrophoresis in the presence of sodiumdodecyl sulfate (13) instead of urea, the procedure was as de-

t To whom reprint requests should be addressed.

4350

The publication costs of this article were defrayed in part by pagecharge payment. This article must therefore be hereby marked "ad-vertisement" in accordance with 18 U. S. C. §1734 solely to indicatethis fact.

Proc. Natl. Acad. Sci. USA 76 (1979) 4351

1 ~~~~+

2

0 3 -"A0

r0 1~~~0

1 0

0 10

0 *I* I0

A I

F/

6

FIG. 1. Assembly for electrophoretic blotting procedure. 1,Electrodes of destainer; 2, elastic bands; 3, disposable pipette-tip tray;4, nitrocellulose sheets; 5, polyacrylamide gel; 6, Scotch-Brite pads.Assembly parts are shown separated for visualization only.

scribed above except that the polarity of the electrodes was

reversed and the electrode buffer was 25 mM Tris-192 mMglycine/20% (vol/vol) methanol at pH 8.3.

Staining for Protein. The blot may be stained with amidoblack (0.1% in 45% methanol/10% acetic acid) and destainedwith 90% methanol/2% acetic acid (see ref. 14).

Immunological Detection of Proteins on Nitrocellulose.The electrophoretic blots (usually not stained with amido black)were soaked in 3% bovine serum albumin in saline (0.9%NaCl/10 mM Tris-HCl, pH 7.4) for 1 hr at 40'C to saturateadditional protein binding sites. They were rinsed in saline andincubated with antiserum appropriately diluted into 3% bovineserum albumin in saline also containing carrier serum withconcentration and species as indicated in the legends. The sheetswere washed in saline (about five changes during 30 min, total)and incubated with the second (indicator) antibody directedagainst the immunoglobulins of the first antiserum. As indicatorantibodies we used '25I-labeled sheep anti-mouse IgG. This hadbeen purified with affinity chromatography on Sepharose-immobilized myeloma proteins and labeled by a modifiedversion of the chloramine T method in 0.5 ml with 0.5 mg ofIgG and 1 mCi of Na'25I (1 Ci = 3.7 X 1010 becquerels) for 60sec at room temperature. The specific activity was approxi-mately 1.5 yCi/Ag of IgG. 125I-Labeled IgG was diluted to 106cpm/ml in saline containing 3% bovine serum albumin and 10%goat serum, and 3 ml of this solution was used for a nitrocellulosesheet of 100 cm2. Incubation was in the presence of 0.01% NaN3for 6 hr at room temperature. The electrophoretic blots werewashed in saline (five changes during 30 min, total) and thor-oughly dried with a hair dryer. The blots were exposed to KodakX-Omat R film for 6 days.

Fluorescein- and horseradish peroxidase-conjugated rabbitanti-goat IgG (Nordic Laboratories, Tilburg, Netherlands) werereconstituted before use according to the manufacturer's in-structions. Fluorescein-conjugated antibodies were used at 1:50

dilution in saline containing 3% bovine serum albumin and 10%rabbit serum. After incubation for 30 min at room temperature,the blots were washed as above and inspected or photographedwith a Polaroid camera under long-wave UV light through ayellow filter.

Horseradish peroxidase-conjugated IgG preparations wereused at 1:2000 dilution in saline containing 3% bovine serumalbumin and 10% rabbit serum. The blots were incubated for2 hr at room temperature and washed as described above. Forthe color reaction (15), the blots were soaked in a solution of 25,.g of o-dianisidine per ml/0.01% H202/10 mM Tris-HCI, pH7.4. This was prepared freshly from stock solutions of 1% o-dianisidine (Fluka) in methanol and 0.30% H202. The reactionwas terminated after 20-30 min by washing with water. Theblots were dried between filter paper. Drying considerablyreduced the background staining. The blots were stored pro-tected from light.

RESULTSElectrophoretic Transfer of Ribosomal Proteins from

Polyacrylamide Gels to Nitrocellulose Sheets. Most proteinsor complexes containing protein adsorb readily to nitrocellulosefilters (16), whereas salts, many small molecules, and RNA areusually not retained. These binding properties are widely usedfor binding assays with nitrocellulose filters. We found thatproteins were retained on these filters equally well when carriedtowards the filter in an electric field. If the electric field wasperpendicular to a slab gel containing separated proteins (seeFig. 1), we obtained a replica of the protein pattern on the ni-trocellulose sheet. This is demonstrated with ribosomal proteinsfrom E. coli; a conventionally stained gel (Fig. 2A) and a stainedelectrophoretic blot of an identical gel (Fig. 2B) are shown. Allribosomal proteins from chicken liver and E. coli ribosomesdetectable on two-dimensional gels could be seen on the elec-trophoretic blots produced from them. An example of a blotfrom a two-dimensional gel is given in Fig. 3. When the originalpolyacrylamide gel was stained after blotting, no protein couldbe detected. Thus, the blotting procedure removed all proteinfrom the gel.To establish whether the proteins removed from the gels were

quantitatively deposited on the nitrocellulose sheet, we sepa-rated 3H-labeled proteins from chicken liver 60S ribosomalsubunit by two-dimensional electrophoresis and compared theradioactivity that could be recovered from the blot with thatrecovered directly from the gel (Table 1). Single proteins orgroups of poorly separated proteins were cut out and radioac-tivity was measured after combustion of the samples. The resultswere within the variability inherent to two-dimensional anal-yses. Variations could be accounted for by variable transfer ofproteins into the second dimension gel and the acuity withwhich spots can be cut out.

At loads exceeding the capacity of nitrocellulose, losses ofprotein occurred. Titration with radioactive ribosomal proteinsunder blotting conditions showed that at concentrations below0.15 ,Lg/mm2 all protein was adsorbed. Overloading becameapparent when a second sheet of nitrocellulose directly un-derneath the first one took up protein or when protein becamevisible on the cathodal surface of amido black-stained blots.The conservation of resolution together with the high re-

covery of ribosomal proteins simplifies the procedure for au-toradiography. The common procedure involving drying ofpolyacrylamide gels under heat and reduced pressure (19),which is tedious and time consuming, may be eliminated. Be-cause the proteins become concentrated on a very thin layer,autoradiography from 14C- and m5S-labeled proteins should behighly efficient even without 2,5-diphenyloxazole impregnation

Biochemistry: Towbin et al.

4352 Biochemistry: Towbin et al.

A

Proc. Nati. Acad. Sci. USA 76 (1979)

B

I~~~~~~~

Fic..Electrophoretic blotting of ribosomal proteins from one-dimensional gels. Total ribosomal proteins from E. coli were separated on

an 18% polyacrylamide slab gel containing 8 M urea. (A) A section of the gel was stained with Coomassie blue; (B) another section was electro-p)horetically blotted and the blot was stained as described in Materials and Methods. Electrophoresis was from left to right.

(19). We have successfully obtained such autoradiograms fromgels of MS-labeled proteins (not shown). Further, preliminaryexperiments with tritiated proteins have shown that dried blotsmay be processed for fluorography by brief soaking in 10%diphenyloxazole in ether (20).The above experiments were done with ribosomal proteins

separated on polyacrylamide gels containing urea. We haveelectrophoretically blotted proteins from sodium dodecyl sulfateby the modified procedure also described in Materials andMethods. Again, there was no loss of resolution. However,differences of staining intensities between proteins on the gel

A

fts

A.

FIG. 3. Electrophoretic blotting of ribosomal proteins fromtwo-dimensional gels. Proteins (35 ,ug) extracted from the 60S ribo-somal subunit of chicken liver (12) were separated by two-dimensionalgel electrophoresis. (A) Gel stained with Coomassie blue; (B) blot ofan identical gel. Electrophoresis: 1st dimension, from left to right(towards cathode); 2nd dimension, top to bottom.

and the blot were apparent. In spite of the apparently incom-plete recovery, blots from polyacrylamide gels containing so-dium dodecyl sulfate may be used for detection of antigen inthe same way as described below for ribosomal proteins (un-published experiments).

Detection of Antigen by Antibody Binding on Blots InSitu. We found that proteins transferred to nitrocellulose sheetsremained there without being exchanged over several days.Because a blot could be saturated with bovine serum albuminto block the residual binding capacity of the sheet, it can betreated as a solid-phase immunoassay. In the following im-munological applications, we used indirect techniquesthroughout. Thus, antibody bound by the immobilized antigenwas detected by a second, labeled antibody directed against thefirst antibody, and in each case excess unbound antibody waswashed out.

Table 1. Efficiency of transfer of ribosomal proteins tonitrocellulose sheets

Protein orgroup ofproteinsanalyzed

Recovery onblot, %

3 1234,4A 1045 1116 1077,8 869 8010 11211 7912, 16 9313 9515, 15A, 18 12517 11519 13921,23 11826 11427 14328, 29 6931 11733 131

Ribosomal large-subunit proteins from chicken liver were tritiatedby reductive methylation (17) and separated by two-dimensionalelectrophoresis (12) in the presence of 35 ,ig of carrier protein. Twoidentical gels were run. One was stained; the other was electropho-retically blotted on a nitrocellulose sheet. Spots were identified ac-cording to our nomenclature for chicken ribosomes (12), which differsonly in minor respects from that established for rat ribosomes (18).Corresponding spots or groups of spots were cut from the gel and theblot. Their radioactivity was determined after conversion to tritiatedwater in a sample oxidizer (Oxymat).

BProc. Natl. Acad. Sci. USA 76 (1979) 4353

A

FIG. 4. Detection of E. coli ribosomal proteins L7 and L12 by (A) horseradish peroxidase- and (B) fluorescein-conjugated antibodies. Totalribosomal proteins from E. coli were separated and blotted as in Fig. 2. The anti-L7/L12 serum had a titer of 340 pmol of 70S ribosomes perml of serum as determined by turbidity formation (20). Incubation was for 2 hr at room temperature in goat antiserum diluted 1:10 in salinecontaining 3% bovine serum albumin and 10%o rabbit carrier serum and then with conjugated anti-goat IgG. In each case the lower strip is a controlwith preimmune antiserum. Electrophoresis was from left to right.

In Fig. 4 the detection of E. coli ribosomal proteins L7 andL12 with a goat serum specific for proteins L7 and L12 isshown. L7 is identical to L12, except for its N-acetylatedNH2-terminal amino acid (21). L7 and L12 fully crossreactimmunologically (22) and are separated on acidic polyacryl-amide gels (21). Both peroxidase- (Fig. 4A) and fluorescein-conjugated (Fig. 4B) antibodies were able to reveal immuno-globulin that was specifically retained by proteins L7 and L12.In each case, the lower gel is a control with preimmune serum.Peroxidase-conjugated antibodies were far more sensitive thanfluorescein-conjugated ones. They could therefore be used atmuch higher dilution. This also permitted the detection of verysmall amounts of antigen. With a rabbit serum (23) we coulddetect 100 pg of L7 and L12 with serum and incubation con-ditions similar to those of the experiment described in Fig. 4 (notshown).

Because we can use the procedure to detect a specific anti-body reacting with a specific protein after electrophoresis inpolacrylamide, we should also be able to determine whichproteins have elicited antibodies in a complex mixture of im-munogens. In the experiment of Fig. 5, individual sera of five

mice immunized with chicken liver ribosomes were tested. Weused 125I-labeled sheep anti-mouse immunoglobulins to detectthe presence of mouse immunoglobulins. In all mice, antibodieswere preferentially produced against slowly moving proteins,presumably of high molecular weight. The procedure can thuscharacterize the antigen population against which specificantibodies have been raised in a mixture of immunogens.

DISCUSSIONThe electrophoretic blotting technique described here producesreplicas of proteins separated on polyacrylamide gels with highfidelity. We obtained quantitative transfer with proteins fromgels containing urea. This was established here with ribosomalproteins. More generally, nitrocellulose membranes have beenused to retain proteins from dilute solutions for their subsequentquantitative determination (16). Still, there remains the possi-bility that certain classes of protein do not bind to nitrocellulose.In this case absorbent sheets other than nitrocellulose or dif-ferent blotting conditions may be helpful.We have demonstrated that proteins immobilized on nitro-

cellulose sheets can be used to detect their respective antibodies.

FIG. 5. Detection of immunoglobulin from individual mice directed against ribosomal proteins from chicken liver. Total protein from chickenliver ribosomes (12) was electrophoretically separated and blotted as in Fig. 2. Sera were obtained from five individual mice immunized againstcombined 40S and 60S subunits. The antisera were diluted 1:50 in saline containing 3% bovine serum albumin and 10% goat carrier serum. Thel)lots were incubated in 250 ,l of the diluted antiserum for 6 hr at room temperature. The blots were combined and treated with 125I-labeledsheep anti-mouse IgG and autoradiographed. Electrophoresis was from left to right.

Biochemistry: Tow.in et al.

4354 Biochemistry: Towbin et al.

With radioactively labeled or peroxidase-conjugated antibodiesthe method is sensitive enough to detect small amounts ofelectrophoretically separated antigen, and this simple procedurecan also be used to show the presence of small amounts ofantibody in a serum of low titer. Because the antigen is immo-bilized on a sheet, the antibody is not required to form a pre-cipitate with the antigen. The blotting technique therefore hasthe potential for immunoelectrophoretic analysis of proteinsby using binding of Fab fragments or binding of antibodiesagainst a single determinant, such as monoclonal antibodiesproduced by hybridomas (24). This could not be done by cur-rent immunoelectrophoretic techniques. If hybridoma clonesare obtained from a mouse immunized with impure immu-nogen, it will be possible to use the technique to screen for clonesmaking antibody directed against a desired antigen. Providedthe desired antigen has a characteristic mobility in polyacryl-amide gel electrophoresis, the appropriate clone can be selectedwithout ever having pure antigen.The procedure described here also has potential as a tool for

screening pathological sera containing auto-antibodies-e.g.,those against ribosomes (25-27). The precise identification ofthe immunogenic components may be a useful diagnostic toolfor various pathological conditions.A further advantage of immobilization of proteins on nitro-

cellulose is the ease of processing for autoradiography. Con-ventional staining, destaining, and drying of polyacrylamidegels takes many hours, and the exact drying conditions are ex-tremely critical, especially for 18% gels as used in the seconddimension for ribosomal proteins (12). When the proteins aretransferred to a nitrocellulose support, as described here, theelectrophoretic blotting takes 1 hr, staining and destaining lessthan 10 min, and drying an additional 5 min. This is thus bothfaster and simpler than conventional procedures, and it elimi-nates the tedious and hazardous procedure of soaking the gelsin diphenyloxazole (19).The technique has been developed to detect specific antisera

against ribosomal proteins. However, it is applicable to anyanalytical procedure depending on formation of a protein-ligand complex. With the blotting technique, the usual proce-dure of forming a complex in solution and retaining it on amembrane would have to be reversed: the protein, alreadyadsorbed to the membrane, would have to retain the ligandfrom a solution into which the membrane is immersed. Inter-actions that can possibly be analyzed in this way include hor-mone-receptor, cyclic AMP-receptor, and protein-nucleic acidinteractions. The ligand may also be a protein. Enzymes sepa-rated on polyacrylamide gels could also be conveniently lo-calized on blots by in situ assays. A critical requirement forthese applications is that the protein is not damaged by theadsorption process and that binding sites remain accessible toligands and substrates. In this respect, considerations similarto those in affinity chromatography and insoluble enzymetechniques pertain.The method could also be adapted to the procedure of

Cleveland et al. (28) for the analysis of proteins eluted frombands in polyacrylamide gels by one-dimensional fingerprints:one could label by iodination in situ' on the nitrocellulose andthen carry out the proteolytic digestion.

In preliminary experiments we have attempted to identifyribosomal RNA binding proteins by binding RNA to ribosomalproteins immobilized on nitrocellulose by the procedure of thispaper, followed by staining for RNA (unpublished data), andhave found a tendency for nonspecific binding. However, J.Steinberg, H. Weintraub, and U. K. Laemmli (personal com-munication) have independently developed a similar procedurefor identifying DNA binding proteins.

We thank Drs. J. Schmidt and F. Dietrich for advice and help withimmunization procedures and Mrs. M. Towbin for advice on settingup the peroxidase assay.

1. Gordon, A. H. (1971) in Laboratory Techniques in Biochemistryand Molecular Biology, eds. Work, T. S. & Work, E. (North-Holland, Amsterdam), p. 62.

2. Williamson, A. R. (1971) Eur. J. Immunol. 1, 390-394.3. Grabar, P. & Williams, C. A. (1955) Biochim. Biophys. Acta 17,

67-74.4. Laurell, C.-B. (1965) Anal. Biochem. 10, 358-361.5. Zubke, W., Stadler, H., Ehrlich, R., Stoffler, G., Wittmann, H.

G. & Apirion, D. (1977) Mol. Gen. Genet. 158, 129-139.6. Showe, M. K., Isobe, E. & Onorato, L. (1970) J. Mol. Biol. 107,

55-69.7. Southern, E. M. (1975) J. Mol. Biol. 98, 503-517.8. Broome, S. & Gilbert, W. (1978) Proc. Natl. Acad. Sci. USA 75,

2746-2749.9. Sharon, J., Morrison, S. L. & Kabat, E. A. (1979) Proc. Natl. Acad.

Sci. USA 76, 1420-1424.10. Hamel, E., Koka, M. & Nakamoto, T. (1972) J. Biol. Chem. 247,

805-814.11. Moller, W., Groene, A., Terhorst, C. & Amons, R. (1972) Eur. J.

Biochem. 25, 5-12.12. Ramjoue, -H.-P. R. & Gordon, J. (1977) J. Biol. Chem. 252,

9065-9070.13. Laemmli, U. K. (1970) Nature (London) 227,680-685.14. Schaffner, W. & Weissmann, C. (1973) Anal. Biochem. 56,

502-514.15. Avrameas, S. & Guilbert, B. (1971 Eur. J. Immunol. 1, 394-

396.16. Kuno, H. & Kihara, H. K. (1967) Nature (London) 215, 974-

975.17. Moore, G. & Crichton, R. R. (1974) Biochem. J. 143, 604-612.18. McConkey, E. H., Bielka, H., Gordon, J., Lastick, S. M., Lin, A.,

Ogata, K., Reboud, J.-P., Traugh, J. A., Traut, R. R., Warner, J.R., Welfle, H. & Wool, I. G. (1979) Mol. Gen. Genet. 169,1-6.

19. Bonner, W. M. & Laskey, R. L. (1974) Eur. J. Biochem. 46,83-88.

20. Randerath, K. (1970) Anal. Biochem. 34, 188-205.21. Terhorst, C., Wittmann-Liebold, B. & Moller, W. (1972) Eur.

J. Biochem. 25, 13-19.22. Stoffler, G. & Wittmann, H. G. (1971) J. Mol. Biol. 62, 407-

409.23. Howard, G., Smith, R. L. & Gordon, J. (1976) J. Mol. Biol. 106,

623-637.24. Kohler, G. & Milstein, C. (1976) Eur. J. Immunol. 6,511-519.25. Schur, P. H., Moroz, L. A. & Kunkel, H. G. (1967) Immuno-

chemistry 4, 447-453.26. Miyachi, K. & Tan, E. M. (1979) Arthritis Rheum. 22, 87-93.27. Gerber, M. A., Shapiro, J. M., Smith, H., Jr., Lebewohl, 0. &

Schaffner, F. (1979) Gastroenterology 76, 139-143.28. Cleveland, D. W., Fischer, S. G., Kirschner, M. W. & Laemmli,

U. K. (1977) J. Biol. Chem. 252, 1102-1106.

Proc. Natl. Acad. Sci. USA 76 (1979)